Laserfiche WebLink
<br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br />I <br /> <br />The Log Pearson III distribution replicates the 2-year return period well. but is below NOAA <br />Atlas for the rest of the data. The Fisher-Tippet distribution done by WRC appears to be only <br />slightly higher with respect to the 2-year return period, and just over 0.30 inches lower for the <br />100-year return period. For the values between the 5- and 25-year return periods, the <br />differences are insignificant. Both of these distributions appear to fit the data well, and the <br />Fisher-Tippet distribution appears to replicate the NOAA Atlas Fisher-Tippet distribution with <br />exception to the 50-, and 100-year return periods. These differences are most likely due to <br />the differences in analyses, mainly the difference between site and regional. Taking this into <br />account. the next step is to do the same distribution with the full range of data from 1900 to <br />1997. <br /> <br />From this analysis, it appears the Log Pearson III distribution applies more emphasis on the <br /> <br /> <br />more extreme values and therefore predicts higher precipitation values for most all return <br /> <br /> <br />periods. The Fisher-Tippet distribution seems to be less effected by the higher precipitation <br /> <br /> <br />values. Because of this, the 2-, 5-, and lO-year return periods remain essentially unchanged <br /> <br /> <br />and the 25-, 50-, and 100-year return periods actually decrease from the first analysis done by <br /> <br /> <br />WRC. The values from the NOAA/NWS preliminary study tend to replicate the results from <br /> <br /> <br />the Fisher-Tippet distribution well. <br /> <br />This seems reasonable if you take into account the length of record used. By using <br />approximately 70 years of record as NOAA Atlas 2 originally did. extrapolation is required <br />to approximate the 100-year return interval. The regional analysis that NOAA Atlas used will <br />add more data points to the data set, but it will also generalize the data used for that region. <br />and in essence decrease the magnitude of the values that are not close to the average for that <br />region. This becomes a trade off. More accuracy is achieved with the regional analysis, but <br />that accuracy corresponds to the average for that region. not necessarily to the individual area <br />that is desired. By using 98 years of data and an "at-site" analysis. the distribution should be <br />much more specific to that area and approximate the l00-year return period more accurately. <br />Therefore. although the Fisher-Tippet distribution tends to support the analysis done by <br />NOAA/NWS, the Log Pearson III distribution appears to fit the data set better (Appendix C). <br />This suggests that NOAA Atlas. when revised, may underestimate the higher return period <br />events. <br /> <br />-6- <br />